My previous blog on “Model Thinking – The Fox and the Hedgehog” discussed the differences between the Hedgehog’s and the Fox’s style of cognitive thinking and decision making. In this blog I am going to talk a little more about why we should study models. In upcoming blogs I will talk about sorting vs. peer effect model and also 2x2 model which helps with decision making during issues involving competing choices.
So why study models?
We study models to become more intelligent citizens of the world, understand data and patterns around us, and get our logic straight during a decision process. We watch the world around us and try to make sense of it. As human beings we are wired to rationalize everything and find patterns. I highly recommend reading the book “Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets” by Nassim Nicholas Taleb. Taleb is a Philosopher who has spent the better part of his post investment banking career and life studying problems of uncertainty, probability, and knowledge. His three books: Antifragile, The Black Swan, and Fooled by Randomness are must reads for leaders and decision makers who have to deal with real life decisions under uncertainty in their day to day lives. The book argues that luck is often mistaken for skill, but I have my own opinion about it, which I will reserve for a future blog.
The human brain abhors randomness. The key point of the book is that the patterns seen on charts or in life is nothing more than figments of our own imagination. It is called Apophenia, which is seeing patterns in random or meaningless data. Humans have the tendency to seek random patterns within information in general, such as with gambling, love, and religion. This argument might seem in contradiction to Machine Learning and the Big Data Analysis theory which is about finding patterns in large data, but it is in fact not. Machine learning algorithms find patterns when they exist, which makes these valid. For example the price of the house is linearly proportional to the square feet of the lot and construction. You can build a mathematical model or equation based model to find the correlation and hence the coefficients of linearity.
Our brains are wired to recognize patterns, whether they exist or not, like watching shapes form in the clouds which is also referred to as Pareidolia. We always feel the need to make sense of our surroundings and events around us. Sometimes, an ability to discern a pattern is very useful, but other times we simply fool ourselves into seeing patterns where there are none. We just can’t deal with arbitrary events. Perhaps, this is one of the reasons why quantum physics – random at its core – is so hard to understand. It is difficult for us to accept that at the core of our reality, there is nothing but uncertainty. I will write about a simple model called “Game of Life”, a one dimensional cellular automata model and also cover the “IT from BIT” theory in future blogs which will try to explain why it is so hard to infer what is going on at the micro level by looking at the macro level, and how complicated micro decisions can get in aggregation.
Even Einstein, one of the founding fathers of quantum physics, could not accept it. He famously said to Neils Bohr, “God does not play dice!” To which Bohr retorted, “Quit telling God what to do!” In this blog I am going to discuss that models help us make sense of the world around us and that we can even build models which can help us understand the patterns.
There is famous quote by George Box where he says, “Essentially all models are wrong, but some are useful”. Models do not give you nicely packaged answers for people who are looking for finality. They are the language of business, economics, academics, philosophy, psychology, and everyday life. Models help you be better at whatever you choose to do. Different models are applicable in different contexts. For example Game Theory is the study of strategic behavior between individuals, companies, and countries.
Models help us become intelligent citizens of the world. To understand segregation between high and low income groups or segregation between different races in parts of cities, it is important to understand models. For you to become involved in a conversation, it is important that you can use and understand models, because models tie us to the mast of logic. Formal models are better at both calibration (how accurate the model is) and discrimination (how fine the predictions are – instead of just saying cold or hot, predict whether it is 80 degrees or 90 degrees).
Models make us think clearly. In any complex event like gambling, stock trading, or horse racing people who use models do better. Models weed out logical inconsistencies and help us think about the consequences of our actions.
Models also help us understand data. There is, what data scientists’ call, a hair ball of data, of which there is no way our mind can make sense of it. There is no way to untangle patterns from this data without models.
Models can help you understand why a certain movie in a theater receives a standing ovation, while others don’t. They help you understand the Arab Spring revolution which overthrew dictators in Egypt and Libya. They help you understand economic growth, and peer effects (why a group of people who hang out together generally look alike, act alike and think alike). They help you understand how people around you affect you (You change your behavior to match others around you), and as in the case of Colonel Blotto’s model it helps us decide how many resources to allocate across different fronts in a war.
Models are built for one purpose and we can apply them to many other purposes to help us become more engaged people out there in the world. When we construct models, we get pretty interesting unexpected results.
There are two types of models.
1. Equation based models or linear models (e.g: y=ax+b). These are easy to model and simple ML algorithms can help find patterns and analyze causality.
2. Agent based model: These models have three parts:
· A bunch of agents – people, firms, countries, organizations which are called objects of the models
· Agents have certain behaviors or they follow certain rules. These rules could be optimal rules (rational choice model - Individuals are doing optimal things in the given context) or irrational rules (where they are not optimizing, but rather following simple rules).
· Then finally, the outcomes. What kind of outcomes can we get? We may expect agents following certain behaviors to have correlated outcomes, but when you work through the logic, the opposite is true sometimes.
A blend of formal models and experience is what gives the best results. Smart people use models, but models don’t tell them what to do. Models make us humble. If you lay out all the logic we realize we had no idea what was going to happen. They help us see the full dimensionality of the problem. People with lots of formal models do better than people with one formal mode. If you want to be out there helping change the world in useful ways, it is really helpful to have some understanding of models. We have a moral obligation to leave a better world for the mankind after us than the one which was handed to us.
So why study models?
We study models to become more intelligent citizens of the world, understand data and patterns around us, and get our logic straight during a decision process. We watch the world around us and try to make sense of it. As human beings we are wired to rationalize everything and find patterns. I highly recommend reading the book “Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets” by Nassim Nicholas Taleb. Taleb is a Philosopher who has spent the better part of his post investment banking career and life studying problems of uncertainty, probability, and knowledge. His three books: Antifragile, The Black Swan, and Fooled by Randomness are must reads for leaders and decision makers who have to deal with real life decisions under uncertainty in their day to day lives. The book argues that luck is often mistaken for skill, but I have my own opinion about it, which I will reserve for a future blog.
The human brain abhors randomness. The key point of the book is that the patterns seen on charts or in life is nothing more than figments of our own imagination. It is called Apophenia, which is seeing patterns in random or meaningless data. Humans have the tendency to seek random patterns within information in general, such as with gambling, love, and religion. This argument might seem in contradiction to Machine Learning and the Big Data Analysis theory which is about finding patterns in large data, but it is in fact not. Machine learning algorithms find patterns when they exist, which makes these valid. For example the price of the house is linearly proportional to the square feet of the lot and construction. You can build a mathematical model or equation based model to find the correlation and hence the coefficients of linearity.
Our brains are wired to recognize patterns, whether they exist or not, like watching shapes form in the clouds which is also referred to as Pareidolia. We always feel the need to make sense of our surroundings and events around us. Sometimes, an ability to discern a pattern is very useful, but other times we simply fool ourselves into seeing patterns where there are none. We just can’t deal with arbitrary events. Perhaps, this is one of the reasons why quantum physics – random at its core – is so hard to understand. It is difficult for us to accept that at the core of our reality, there is nothing but uncertainty. I will write about a simple model called “Game of Life”, a one dimensional cellular automata model and also cover the “IT from BIT” theory in future blogs which will try to explain why it is so hard to infer what is going on at the micro level by looking at the macro level, and how complicated micro decisions can get in aggregation.
Even Einstein, one of the founding fathers of quantum physics, could not accept it. He famously said to Neils Bohr, “God does not play dice!” To which Bohr retorted, “Quit telling God what to do!” In this blog I am going to discuss that models help us make sense of the world around us and that we can even build models which can help us understand the patterns.
There is famous quote by George Box where he says, “Essentially all models are wrong, but some are useful”. Models do not give you nicely packaged answers for people who are looking for finality. They are the language of business, economics, academics, philosophy, psychology, and everyday life. Models help you be better at whatever you choose to do. Different models are applicable in different contexts. For example Game Theory is the study of strategic behavior between individuals, companies, and countries.
Models help us become intelligent citizens of the world. To understand segregation between high and low income groups or segregation between different races in parts of cities, it is important to understand models. For you to become involved in a conversation, it is important that you can use and understand models, because models tie us to the mast of logic. Formal models are better at both calibration (how accurate the model is) and discrimination (how fine the predictions are – instead of just saying cold or hot, predict whether it is 80 degrees or 90 degrees).
Models make us think clearly. In any complex event like gambling, stock trading, or horse racing people who use models do better. Models weed out logical inconsistencies and help us think about the consequences of our actions.
Models also help us understand data. There is, what data scientists’ call, a hair ball of data, of which there is no way our mind can make sense of it. There is no way to untangle patterns from this data without models.
Models can help you understand why a certain movie in a theater receives a standing ovation, while others don’t. They help you understand the Arab Spring revolution which overthrew dictators in Egypt and Libya. They help you understand economic growth, and peer effects (why a group of people who hang out together generally look alike, act alike and think alike). They help you understand how people around you affect you (You change your behavior to match others around you), and as in the case of Colonel Blotto’s model it helps us decide how many resources to allocate across different fronts in a war.
Models are built for one purpose and we can apply them to many other purposes to help us become more engaged people out there in the world. When we construct models, we get pretty interesting unexpected results.
There are two types of models.
1. Equation based models or linear models (e.g: y=ax+b). These are easy to model and simple ML algorithms can help find patterns and analyze causality.
2. Agent based model: These models have three parts:
· A bunch of agents – people, firms, countries, organizations which are called objects of the models
· Agents have certain behaviors or they follow certain rules. These rules could be optimal rules (rational choice model - Individuals are doing optimal things in the given context) or irrational rules (where they are not optimizing, but rather following simple rules).
· Then finally, the outcomes. What kind of outcomes can we get? We may expect agents following certain behaviors to have correlated outcomes, but when you work through the logic, the opposite is true sometimes.
A blend of formal models and experience is what gives the best results. Smart people use models, but models don’t tell them what to do. Models make us humble. If you lay out all the logic we realize we had no idea what was going to happen. They help us see the full dimensionality of the problem. People with lots of formal models do better than people with one formal mode. If you want to be out there helping change the world in useful ways, it is really helpful to have some understanding of models. We have a moral obligation to leave a better world for the mankind after us than the one which was handed to us.
No comments:
Post a Comment